Effects of ZNF804A on neurophysiologic measures of cognitive control

Schizophrenia has heritability estimates of 70–85%¹ and genetic risk in most cases is presumably a result of multiple genetic loci of small effect. In recent years, genome-wide association studies have facilitated the discovery of potential risk genes for neuropsychiatric disorders. ZNF804A rs1344706 was the first single-nucleotide polymorphism (A/C, A = risk allele) to show a significant association with schizophrenia in genome-wide association studies (P = 1.6 × 10^–7).² Although the physiological function of ZNF804A and its role in schizophrenia risk is currently unclear, the gene has been shown to be associated with psychosis,³ and several potential intermediate biological phenotypes, such as deficits in executive function⁴ and altered cortical connectivity during working memory^5,6 and theory of mind.⁷

Cognitive deficits in schizophrenia are broad and conspicuously include deficits in cognitive control, an executive function that refers to the ability to direct behavior toward a goal in the presence of conflict. Studies have shown altered function and connectivity of the dorsolateral prefrontal cortex (DLPFC) and anterior cingulate cortex (ACC) in patients with schizophrenia during cognitive control tasks.⁸ Recently, these alterations have been suggested to represent an intermediate phenotype because they are also found in healthy siblings of patients with schizophrenia.^9,10 In this study, we aimed to explore the role of ZNF804A rs1344706 on activation and functional connectivity measures related to cognitive control. To this end, we used the Modified Flanker task that includes response inhibition (RI) and interference monitoring and suppression (IMS), two important components of cognitive control function. A total of 208 Caucasian healthy volunteers with no history of psychosis and no first-degree relatives with a schizophrenia spectrum disorder and with good quality functional magnetic resonance imaging (fMRI) data and genotype information for ZNF804A rs1344706 were selected from the larger sample of the Clinical Brain Disorders Branch Sibling Study of schizophrenia at the National Institute of Mental Health (DR Weinberger PI). In all, 89 subjects were rs1344706 AA homozygotes, 93 were rs1344706 AC heterozygotes and 26 were rs1344706 CC homozygotes. The genotype groups did not differ across demographic and task performance variables based on ANOVA and chi-square analyses (see Supplementary Materials, Supplementary Table 1).

Subjects underwent BOLD fMRI using a gradient echo-planar imaging sequence on a 3T GE Signa scanner during which they performed a modified Flanker Task. fMRI data was processed using SPM5 (SPM; http://www.fil.ion.ucl.ac.uk). Genotype effects on cognitive control-dependent neural response and connectivity were tested using random-effects general linear model ANOVA analysis (P<0.05; small volume corrected P_fdr<0.05) (see Supplementary Materials). To examine the cognitive control-dependent modulation of functional coupling of ACC and DLPFC, a psychophysiological interaction (PPI) analysis was performed using both ACC and right DLPFC as seed regions. The choice of ACC (identified through WFU-pickatlas toolbox; http://fmri.wfubmc.edu/software/PickAtlas) as one of the seed regions was based on an earlier observation of altered ACC/DLPFC coupling during cognitive control representing a potential intermediate phenotype.⁹ The choice of the second seed region in the right DLPFC (right Brodmann areas 9 and 46) was based on results from previous studies,^5,6 which showed that the coupling of this brain region with other brain regions is modulated by ZNF804A rs1344706 during cognitive processing (see Supplementary Materials).

During RI, although there was no significant difference in activation across genotype groups, PPI analysis revealed increased connectivity between the ACC and right DLPFC in A, risk allele, homozygotes as well as heterozygotes when compared with C homozygotes. This was observed irrespective of the seed region, the right DLPFC (x, y, z = 6, 27, 21, Z = 3.73, P_fdr = 0.037, see Figure 1a) or ACC, although the latter did not survive correction for multiple comparisons (P = 0.006 uncorrected). There were no other areas that showed a significant effect of genotype on functional coupling with the seed regions during RI. During IMS, A, risk allele, homozygotes as well as heterozygotes showed decreased right DLPFC and ACC activation when compared with C homozygotes (DLPFC: x, y, z = 51, 39, 15, Z = 4.00, P_fdr = 0.011; ACC: x, y, z = 3, 30, 27, Z = 2.79, P_fdr = 0.026; see Figure 1b). PPI revealed no significant difference in connectivity across genotype groups during IMS.

Figure 1.

Effect of ZNF804A rs1344706 genotype on cognitive-control-related neural function during RI and IMS conditions of the Modified Flanker task. (a) During RI, PPI analysis revealed increased rDLPFC connectivity with the ACC (P_fdr = 0.037) in A, risk allele, homozygotes when compared with C homozygotes. For illustrative purposes, parameter estimates (in arbitrary units with 90% confidence interval (CI) of ACC-PFC connectivity from the peak voxel in ACC are displayed in the bar graph on the right for the three genotype groups. (b) During IMS, A, risk allele, homozygotes and heterozygotes showed decreased right dorsolateral PFC (P_fdr = 0.011) (b, top image on left) and ACC (P_fdr = 0.026) (b, bottom image on left) activation when compared with C homozygotes. For illustrative purposes, parameter estimates (in arbitrary units with 90% CI) of BOLD signal change from the peak voxel in the right dorsolateral PFC (b, bar graph on top right) and ACC (b, bar graph on bottom right) are displayed in the bar graph on the right for the three genotype groups.

Our results show that ZNF804A modulates mechanisms underlying cognitive control. The ZNF804A rs1344706 allele load effect on ACC-PFC coupling, with risk allele carriers showing increased coupling, adds to evidence that ZNF804A modulates cortical network connectivity during executive cognition.⁵ We have previously found that this pattern of enhanced ACC-PFC functional coupling is associated with schizophrenia and with increased genetic risk for schizophrenia,⁹ suggesting that it is an intermediate biological phenotype related to the genetic risk architecture of illness. This adds to evidence that ZNF804a may confer risk for schizophrenia by impacting this intermediate phenotype mechanism.

Interestingly, we also found that ZNF804A modulates ACC and DLPFC activation during IMS. Although a similar alteration in ACC and DLPFC activation during IMS was observed in patients with SCZ,⁹ it has not been shown to be an intermediate phenotype related to risk for schizophrenia. Therefore, further studies may be necessary to clarify if this effect of ZNF804A on ACC and DLPFC activation during IMS is independent of the mechanism through which it confers genetic risk for schizophrenia.